GRAPH ANOMALY DETECTION: A SURVEY

Summary from the TLDR AI newsletter

This paper offers an in-depth review of deep learning methods, particularly graph neural networks (GNNs), for detecting anomalies in graph data. Researchers introduce a novel taxonomy of approaches, analyzing GNN designs, proxy tasks, and anomaly measures.